The concept of response generalization was examined by means of paired-association learning. Subjects learned two or three lists successively, each list being composed of several pairs of nonsense syllables (stimuli) and three-syllable verbs (response). The stimulus-syllablesin the second (and third) lists were the same as those in the first list. But response-words in eachlist were different from those in the other lists, being similar, opposite or neutral in meaning. Each pair was presented by a memory drum. The anticipation method was adopted. In Exper. 1, the generalization between similar words was studied. Nine subjects worked on two lists successively, each containing six pairs of stimulus and response-words, until one perfect trail was achieved. They recalled and relearned the first list under retroactive conditions, the second list in proactive conditions. Under rest conditions only one list was used. In each condition the time interval between the learning and recall was 5 min., 1 hr. or 24 hrs. In Exper. 2, the meaning relations between the two response-words were similarity, opposition and neutrality. Two lists, each containing 10 paired words, were successively learned to the degree of seven-tenths, and recalled after 30 sec. Then they were relearned immediately to the same degree. Rest conditions were added. Twelve subjects experienced all the four condition (i.e. similarity, opposition, neutrality and rest). In Exper. 3, the meaning relations examined were the same as in Exper. 2, but three lists instead of two were learned. Fifteen subjects were divided into three equal-numbered groups, each of which worked under one set of conditions only. Under S conditions three response words (R1, R2, and R3) were similar to each other. Under O conditions, R1 had oppodite to R2 and R3, which therefore were similar. Under N conditions all the response-words were neutral to each other. Rest conditions were omitted. The time interval between learning and relearning was 24 hrs. The following results and conclusions were obtained : 1) In the process of learning two or three similar or opposite response-words there appeared to be evidence of positive transfer. Even where it did not appear, the learning proceeded at nearly the same speed as the learning of neutral response-words. 2) Recall and relearning under conditions S and O were, for the most part, better than under conditions N and R. In no instance were they worse. These results indicate that positive tranfer and retroacitve facilitation can be derived from a greater degree of generalization, because the generalization between similar or opposite response-words is degree than that between neutral response-words. This conclusion agree partly with Morgan and Underwood's assumption of “parasitic reinforcement.” But it denies E. J. Gibson's contention that negative transfer and retroactive inhibition arise from greater generalization. It also denies Osgood's assertion that opposite relation generates more negative transfer and retroactive inhibition than similar and neutral relations. In short, it seems to us that our results suggest the facilitative function of generalization in verbal learning. 3) Analyzing closely the intrusions in learning and relearning, we found: (a) that generalization is accompanied by differentiation at least in verbal learning; (b) that generalization along the dimension of meaning must be distinguished from one arising from other factors; and c that generalization between opposite words is of a diferent character from the generalization between similar words.
From a logical point of view, social learning, in which a follower rat learn response in the presence of a leader rat responding to a cue, can be regarded as a type of discrimination learning. Therefore, in order to test the continuity controversy over the role of reinforcement, a reversal experiment, combined with social learning experiment, was conducted. Each of 45 rats was run after a leader rat 8 times a day, upon a simple T maze, after 24 hours of food deprivation. In the 1st experiment (Fig. 2-5), rats were rewarded with food for their right responses, but were punished with mild electric shock for their responses to the irrelevant cue; in the 2nd experiment (Fig. 6-8) they were not punished but only withheld food reward for their incorrect response. The experimental situation is shown in Fig. 1. The rats were divided into 6 kinds of groups, the required responses of which were the following: (1) non-imitative, positional response (Fig. 2); (2) non-imitative, discrimination of the lamp, without reversal training (Fig. 3, group II, and Fig. 6); (3) imitative response, without reversal training (Fig. 4, Group II, and Fig. 7); (4) inverse-imitative response (Fig. 5 and Fig. 8); (5) non-imitative, discrimination of the lamp, with reversal training (Fig. 3. group V); (6) imitative response, with reversal training (Fig. 4. group VI) Respective cues to the right responses were (1) kinaesthetic or spacial orientation; (2) and (5) a lamp switched on; (3), (4) and (6) the leader's response to positional or lamp cues. The criterion of the problem solution was defined as the achievment of the performance, in which the rat shows no longer as many as 25% errors. Taking the total number of trails before the completion of learning as an index of the learning speed of the rat, a comparison of the performances of the above groups gives the following results: 1) in the presence of the leader rat, a positional response irrespective of the leader rat was most easily acquired. 2) in the presence of the leader rat, discrimination of the lamp was comparatively easily acquird, and the behavior of the non-imitator was stable. 3) imitation of the leader rat was apparent, but the behavior of the imitator was not so stable as that of the non-imitator. 4) the inverse-imitation of the behavior of the leader rat was almost unseccessful. 5) the retardating effect of the reversal training was not perceived in the groups of both imitative and lamp discrimination.
1. The main problems of the present experiment are to investigate (1) the functions of irrelevant or static stimulus S3 and secondary reinforcing stimulus S4, and (2) the generalization of inhibition to original stimulus S1 through the differential reinforcement by S1 and S2. 2. Experiment (1). Conditioned bar-pressing response to food was established to the presentation conditioned light stimulus S1. And then the differential process by S1 (0j. n. d) and S2 (2j. n. d. or 4j. n. d.) was carried out. The condition to the responses to differential light stimulus S2 were as follows. Group A (non-reinforcement but with the sound of food-release mechanism S4), Group B (non-reinforcement and without S4), Group C (non-reinforcement but with the weak electric stimulus and S4), Group D (non-reinforcement but with the weak electric stimulus and without S4) All group were given twenty differential trials by S1 and S2, after 100 reinforcements to conditioned S1……R responses. Light stimuli S1 and S2 were presented ten times each in random alternative order. Then the resistance of conditioned generalized responses to S was measured. But Group A0 was extinguished immediately after the conditioning process, and not given the differential trails. 3. Experiment (2) After the 50 reinforcements were given to conditioned response (S1……R), differential process by S1 and S2 was carried out as in the above experiment, and then S1……R was extinguished. Control Group was extinguished without interposition of differential process. In this experiment, the conditions of Group B and D in Experiment (1) were used as the S2 conditions. The influence of the differential process to S1……R was measured by the strength of resistance to extinction. 4. Experiment (3) Cionditioned bar-pressing response were reinforced 50 times under the continuous presentation of light stimulus S3. Subjects were divided into two groups: Group A was extinguished by the ordinary extinction procedure, Group B was extinguished under the same conditions of Group A although the stimulus S3 was never presented it extinction process. 5. Results and interpretation. a. S1…R connection was superior in the strength of resistance to extinction. And as the distance from S1 to S2 in stimulus continuum increased, the strength of resistance to extinction of S2……R decreased. (Table 2, 3) b. Without S3 and S4 in the situation of extinction, the extinction process occurred more rapidly. (Table 4) c. Under thedifferential process, S1ER gradually decreased and then was recovered to somewaht below the maximum M. The amount of decrements of S1ER was determined by the condition of S2. (Figs. 2, 3, 4, 5) The above results show that S3 is neutralized through the differential process, and that S1IR generalized to S1.
One of the most outstanding characteristics of GSR is adaptation, or diminution of the response upon repeated stimulation. There are, it must be added certain experimental designs in which the investigators are obliged to administer a stimulus repeatedly to make their investigations. Therefore the investigators who utilize GSR for any purpose and very often required to take into their consideration the effect of repeated stimulation upon the response. The present studies were undertaken to investigate some aspects of this fact and divided into two parts: 1) Selection of the appropriate unit for use in the measurement of GSP, especially in the situation where repeated stimulation is employed. The investigation ig the appropriate unit was most recently undertaken by Lacey, Haggard et al., who successfully selected conductance as the most proper unit-the same conclusion reached in the physiological observation by Darrow some ten years ago, but those researches did not cover the occasion above mentioned. Then our problem is to examine whether the unit selected by Lecey et al. and other conceivable units can be adopted in our case. The raw data were obtained by giving each subject a loud noise ten times and then converted to several sorts of the conceivable units including conductance proposed by Lecey et al. The results lead to the following conclusion. As for the basal level, conductance is the most desirable unit from the viewpoint of normality of distribution (Table 1). As for GSR, change in conductance, % change in conductance and % change in resistance are the most acceptable units in view of both normality of distribution and independence of the basal level (Table 1, Table 2). 2) Analysis of the nature of GSR's diminution. A slight observation could make clear that diminution of the response has many aspects. First, individual differences may attract observers' attention. Another aspect is that the variabilities of the response are partiallu due to the kind of stimulus used. For example, if an investigator employed a series of words as stimuli, he would easily be convinced that every word has a different effect upon the response of his subjects. Moreover, if he changed the position of words in the series, he might observe that the changed position produces a different response. Thus diminution of GSR is complicated by many factors en bloc, among which these three are perhaps the most important ones. To analyse the nature of diminution, then, experimental designs in statistics are required. Our investigation in this part was to ascertain this fact statistically in hope of marking a step to make further progress on the studies of GSR and included two experiments, of which each was undertaken in different design. In one experiment, each subject was given a loud noise twelve times individually and the data obtained were analysed by Randomized Block Experiment (Table 4). The result shows that difference for diminution and for individuals are also significant. In another, each subject was given a series of twelve words selected from Japanese vocabulary as comparatively indifferent ones and randomized in their position. The data obtaind were analysed by Latin Square Design (Table 6). It can be concluded then that differences for stimuli, for individuals and for position are also significant.
This experiment is a study of the relation between the intellectual congnition or the logical intelligence and the intellectual activity or the practical intelligence of the child. The subjects we used were ten superior children (C. A. 11; 9-12; 6, I. Q. 119-136) and ten inferior ones (C. A. 11; 7-12; 5, I. Q. 72-98) in the 6th grade of the primary school, and ten feeble-minded children (Debility and Inbecility C. A. 10; 2-14; 3, I. Q. 41-66) for the supplementary experiment. These children were told to lead a ball out of a detour-box by the tools. The principal results were as follows: 1. Apprehension of the visual structure of a detour-situation was easy for the superior and inferior children but was considerable difficult for the feeble-minded. The process of thinking there observed from the autonomous habitual reaction to the discovery of a detour-course is regarded as a “dynamic change of figure-groud relief” in the thinking. 2. In the detour-handling of rolling, scooping up, giving jumping motion to the ball there could be found no difference in the results of the three groups. This fact suggests that the intellectual congnition and the intellectual activity have no relation to each other insuch a problem. However, in a situation in which a preliminary cognition before handling determines the success or failure of the act, such as the one in which one fails to pass the ball over a place unless one has stopped a hole with some tool brforehand, the superior children showed a more excellent adaptation than the other groups. In such a case, therefore, we observe that the intellectual cognition positively and functionally affects the intellectual acitivity. 3, “The solution by direct insight” of the total situation was very rare even with the superior children; but they showed a remarkably better results in the “solution by insight after errors”than the other two group. We wish to explain these facts as one of the concrete processes of the “recoganization of the total structure”.
In the study of the visual pricess of perception, it is important that not only the factor of stimulus distribution be taken into consideration. From this viewpoint, the experiment of the figural after-effect proposed by Köhler and Wallach is very interesting but it would seem that it should be retested on the ground of quantification from the neutral stand-point because they imply a bold physiological hypothesis on the basis of qualitative observation. Recently many writers have examined the quantity of after-effect, but many of these studies comprise only partial research of the fact and consequently they can not test the validity of the Köhler hypothesis as a whole. Moreover it is difficult to theoreize about the facts within the results of their investigations systematically because the figures and methods they used were different one from another. Taking these points into consideration, it is the perpose of the present writer to measure the quantity of the figural after-effect and to ascertain the functional principles working there in order to test the contradictory theories that have been fequently suggested. The present writer suggests that this phenome-non can be differentiated into two part, the “displacement effect” and the “size effect”. 1. Experimental study of the Gibson effect. The phenomenon named “Gibson effect” is the earliest discovery and is the most frequently measured phenomenon concerning the “displacement effect” Now the writer proposes a modification of Gibson's method in order to measure the after-effects of a curved line. (1) Gibson suggested a hypothesis about adaptation and after-effects of the prolonged inspection of the curved line. However, his method of measuring adaptation was the same in principle as that of measuring after-effects, and accordingly his adaptation theory may be said to have no factual basis at all. (2) The writer confirmed that the previously exposed curved line affected the subsequently exposed straight line causing it to curve in the opposite direction and that, this effect was produced only by the influence of prolonged inspection of the curved line and not by the measuring operation including the direction of adjustment and the constellation of those figures (Exp. A. B.) (3) with respect to the results of our exp. B in which the curved line (I. F.) gradually changed to assume the farm of a straight line through prologed inspection, Gibson might suggest that this was caused by adaptation process, but the existence of this process could not be confirmed by this sort of experiment only. Köhler and Wallach, etc. explain this phenomenon on the basis of the distance between the inspection line (I. F.) and the test line (T. F.). They do not assume the process of normalization. (4) The present writer confirmed that the curvature changed in the direction of a straight line even in the case when the I. F. coincided with the T. F. (Exp. C. D.). To explain the results of exp. A, B, C, and D. systematically, it would be more convenient to do so interms of normalization hypothesis than in terms of Köhler's theory. (5) When, under the condition of Exp. C. in which the I. F. exactly coincided with the T. F., the curvature of the lines was changed variously, the direction of the displacement of the test line was always the same. (6) Whether the I. F. more curved or less curved than the after-effect of the curved line always produced the decrement of curvature od T. F. (Exp. F, G, H.). These results were in disagreement with Köhker's explanation of Gibson effect based upon the principles of displacement and distance paradox. (7) To test Köhler's hypothesis directly, the author compared the effect of the curved I. F. upon the curved T. F. (Exp. 11) with that of the linear I. F. upon the curved T. F. (Exp. 12).
Purpose The purpose of the present study is to determine figural after-effects (Köhler-effect) and Gibson's negative after-effects quantitatively as a function of the inspection time and of the time after inspection. Method The appratus is shown in Fig 1. Figures used in Exp. I, Exp. II and Exp. IV are shown in Fig. 2, Fig. 7 and Fig. 10 respectively. The general procedure was the same as Köhler's and Wirt's “Vollreihen Mothode” was used to measure the amount of after-effect. The course of the declination of after-effects. The course of the declination of after-effects was caught every five seconds by subjects' judgement. The length of the inspection period were 1, 5, 15, 30, 60, 120 and 240 sec. The test for the declining process of figural after-effects in Exp. III under the same arrangement of figures as Exp. I was performed by Hammers' second method which eliminated the factor of successive induction. Exp. V was the test-experiment performed under the condition shown in Eig. 13 which was considered to be the test situation of Exp. IV. Results 1) Under the same conditions of experiments, the Köhler-effect and the Gibson-effect were demonstrated to be identical process. 2) The amount of after-effects (including Gibson's negativre after-effects) depends upon the size and arrangement of figures. If the size and arrangement of figures are kept constant, the amount of after-effects percieved immediately after the inspection of I. F. remains constant irrespective of the length of the inspection period. On the contrary as regards colour-effects, the longer the inspection time is, the more conspicuous are the effects. 3) The amount of after-effects is maximum immediately after inspection, at first it diminishes quickly, then later gradually. When the inspection time is shorter, the gradient of the decay-curve is sharper. The decay-curves of after-effects strictly agree with Muller's formula. 4) The so-called growth-curves of the after-effects that have been determined by Hammer etc. are shown to be nothing more than thecurve of after-effects considered as a function of the inspection time under the condition of several seconds after the inspection of I. F. Therefore, these curves reveal only the limited case of the growing process of after-effects. 5) The gradient of the decay-curve of after-effects that in the decay-curve under the condition of the larger I. F. (see Fig. 2) is slower than that of the smaller I. F. (see Fig. 7). The longer the inspection time is, the more prominent is the defference of the gradient in decay-curves. 6) Although in determining the decay-process of after-effects the method of successive judgement is a very convenient procedure as Hammer has pointed out, there is a significant difference between the results by the method of successive judgement (Hammer's first method) and those by the method of white-paper-insertion (Hammer's second method) in our experiments. In order to get the correct data, it is neccessry to modify the data obtained by the method of successive judgement. 7) Gibson-effects are influensed more or less by the direction of the curved-line. This fact may be explained by the factor of “Zentrishe Schrumpfung des Sehraumes” (Obonai's theory) and by the factor of the potential-illusion in comparative judgement of parallel curved-line (V. F. and T. F.) 8) In experiment on figural after-effects, especially in studying temporal factors, we must take into consideration the judgement mechanism of the subjects.